This work presents a thorough method for object detection and dehazing in real-time videos using deep learning algorithms. Using techniques such as Pytorch MAP-net for video dehazing and the Dark Channel Prior for picture dehazing, our methodology combines these methods to improve visual clarity in hazy real-time video streams. Modern object recognition models like ssd_mobilenet, Faster_rcnn_resnet, EfficientDet_D4, and Yolov8 are used to improve visibility and allow precise object identification in murky settings. Our technique shows notable increases in dehazing quality and object detection accuracy through thorough review and experimentation, providing useful deployment in surveillance, autonomous driving, and environmental monitoring scenarios. This study presents a useful resource for scholars and practitioners in deep learning-based video analysis, and it advances real-time object detection.